The 8-month arc · May 2025 engagement → CTV Aug 11 2025 → Apr 2026.
The feedback loop was disconnected. The way they measured success was wrong. Acquisition strategies were hitting diminishing returns. Spend climbed +64% MoM yet new-customer CAC blew +55% and one-time CAC blew +142% - Meta CAC held flat at $138 and that was the headline they were tracking. The composed view told the truth: the system needed a strategic pivot.
Operating model reframe delivered. “It was never a Meta problem. It was a mix problem.” Audit → align → bridge. Cash-view (CFO) and accrual-view (CMO) reconciled in one operating model. Vendor measurement became a guide, not a gospel. Triangulation replaced single-source decisioning.
Phase 1 hero · Aug 11 – Oct 5 2025. Halo-forward CAC compressed from a $42.69 pre-CTV baseline to $32.58 over 8 weeks (−24%). Phase 2 (Oct – Dec, 9 weeks of holiday brand-layer cuts) drifted to $41.70. Phase 3 (Dec – Mar, 16 weeks, brand-layer restored at 40% intensity) settled at $40.20. Every week of the 33-week arc held below the pre-CTV baseline. MER trajectory aligns with brand-converted demand landing in highest-retention cohorts.
Is the relationship real?
L2Weekly read. When TV moves, do downstream channels actually follow? Three different math approaches all confirm the same answer here. Not a fluke, not double-counted, holds up to a CFO audit.
When does today's TV spend show up in acquisition MER?
When TV moves, here's what follows
checked two different ways| Channel | Strength · linear | Strength · ranked | Lag to peak | Confidence |
|---|---|---|---|---|
| Stablebrand-layer $ → Reactivations | +0.80 | +0.78 | 2w | High |
| AcqAcquisition MER | +0.93 | +0.91 | 2w | High |
| AcqAcquisition CAC | -0.59 | -0.61 | 2w | High |
| ReactReactivation Subs | +0.96 | +0.95 | 2w | High |
| GoogNew Visits Brand Srch + PMax | +0.89 | +0.86 | 7d | High |
| GoogAcq Rev Brand Srch + PMax | +0.83 | +0.80 | 30d | High |
Solves "every channel claims the lift"
cross-channel credit isolationShow the mathThree different ways of reading the relationship · why we use all three
Linear strength. The standard correlation read. Sensitive to outliers · a few weird weeks can pull the number up or down.
Ranked strength. Same idea but compares the ranking instead of the absolute values. Robust to outliers. If linear and ranked agree, the relationship is real, not driven by a few extreme weeks.
Cross-channel isolation (Ridge regression). Solves the "every channel takes credit" problem. Standard reporting tools double-count: Meta says it drove the lift, Google says it drove the lift, Northbeam says they all drove the lift, and the totals add to 165% credit. Ridge applies a penalty when channels overlap, redistributes credit cleanly to 100%.
When all three agree, the read survives a CFO audit. When they disagree, diagnose the data shape before acting on either number.
Open-source: statsmodels.tsa.stattools.grangercausalitytests · pingouin partial correlations · ridge regression with alpha tuned via cross-validation. All free.